Research And Implementation Of Image Processing Algorithm For Corneal Topography Mapper

In recent years,the younger age and higher incidence of myopia among adolescents has become a key public health and social issue of national concern.The birth of the computer-aided corneal analysis system has brought an important change to ophthalmology auxiliary diagnosis and greatly facilitated the development of related ophthalmic instruments.The corneal topography instrument is based on the Placido disc,which extracts the corneal feature information of the collected Placido disc optical imaging pictures,and evaluates the topography information through quantitative indicators,so as to help clinicians better diagnose and treat human eye optics.lesion.With the rapid development of image processing technology and machine vision imaging technology,research on key intelligent algorithms in corneal topography has become a research hotspot.At present,the mainstream corneal topographs on the market are still mainly manual,mainly relying on the operation of doctors for shooting and collection,and system automation is the direction of future development.By reducing external intervention to reduce errors,it can greatly improve the Enhancing the work efficiency of doctors has important practical significance and clinical value for ophthalmic adjuvant therapy in the clinical medical field.For this reason,this paper focuses on the key algorithms in the instrument system—the improved pupil positioning and tracking algorithm based on the Average of Synthetic Exact Filters in image acquisition,the center positioning of the Placido disc image ring,the ring center positioning in the image post-processing.Edge detection and feature point extraction,and build a preliminary software platform based on MFC(Microsoft Foundation Classes Library).The specific work content is as follows:(1)Real-time positioning and tracking algorithm of pupil based on ASEF improvement.When the traditional ASEF trains the synthetic average filter for matching,each image in the data set will only be used once to form a single image response result,and finally the individual results of all the data set training will be summed and the mean value will be used to suppress the interference information.This cannot efficiently use the feature information of each picture.First of all,this paper studies this problem.By improving the synthetic average filter training method,the target area is determined by the region growing algorithm in the training process,and a suitable adaptive weight filter is constructed through the size of the target area.During training,according to The position of different points in the area gives weights to the response of the point,and finally an improved adaptive synthetic average filter is obtained by the mean value of each image in the data set.The improved algorithm makes full use of the correlation information between each point in the target area and the image,so that the characteristic of determining the filter weight based on the distance is introduced in the process of solving the synthetic average filter,and the feature information of the image can be used more fully.And it is more effective to average and suppress the background interference information.At the same time,the improved algorithm retains the advantages of ASEF’s simple operation of converting time-domain convolution into frequency-domain product,and improves the accuracy of construction filter matching detection.Finally,the performance of the improved algorithm is verified by experiments.In this paper,the image is rotated at different angles such as +10,+5,-5,-10 during training,so that the trained filter can be applied to more complex situations.Experiments show that the accuracy of the new algorithm has been greatly improved in terms of time-consuming detection and accuracy.(2)In this paper,the corneal image collected by the above-mentioned tracking algorithm is processed.Research on the sharpness judgment algorithm in image acquisition,according to the image characteristics in this paper,select the appropriate algorithm for sharpness judgment in order to collect a clearer image.In the image post-processing,the corneal image is pre-processed by filtering,cropping,etc.,and the effective area of the corneal ring is segmented and extracted according to algorithms such as adaptive threshold,corrosion expansion,and connected areas,paving the way for subsequent data post-processing.Finally,the center positioning algorithm of the ring is studied,the center is positioned through the Hough gradient transformation algorithm(Hough)and corrected based on sub-pixels,and the ring is extracted from 24 or 32 rings and the features on the corresponding ring Click to extract.(3)Based on the MFC graphical interface framework,this paper preliminarily builds the software platform design,and realizes the basic functions of the software with the database.Through the above work,the research and implementation of the image processing algorithm for the corneal topographer is basically completed.Through the analysis of the experimental results,the algorithm can better meet the application requirements of the actual lightweight platform.